Academic journal article Economic Quarterly - Federal Reserve Bank of Richmond

The Wealth Effect in Empirical Life-Cycle Aggregate Consumption Equations

Academic journal article Economic Quarterly - Federal Reserve Bank of Richmond

The Wealth Effect in Empirical Life-Cycle Aggregate Consumption Equations

Article excerpt

This article presents an empirical model of U.S. consumer spending that relates consumption to labor income and household wealth. This specification is consistent with the life-cycle hypothesis of saving first popularized in the 1960s by Ando, Modigliani, and their cohorts.1 My analysis here extends the previous research in several directions. First, I examine the dynamic relationship between consumption, income, and wealth using cointegration and error correction methodology. In previous research, the traditional life-cycle model has often been examined using either levels or first differences of these variables. While the use of differences does avoid the pitfall of spurious correlation due to common trending series, it tends to lead to the omission of the long-run equilibrium (cointegrating) relationships that may exist among levels of these variables. In fact, Gali (1990) goes so far as to present a theoretical life-cycle model that generates a common trend in aggregate consumption, labor income, and wealth. Therefore, my empirical work here tests for the presence of a long-run equilibrium (cointegrating) relationship between the level of aggregate consumer spending and its economic determinants such as labor income and wealth. I then examine the short-run dynamic relationship among these variables using an error correction specification proposed in Davidson et al. (1978).

The present article investigates whether wealth has predictive content for future consumption. If it does, then changes in wealth may lead to changes in consumer spending.2 I also examine whether consumer spending is sensitive to stock market wealth. The 1990s saw an enormous increase in household wealth generated by the rising value of household stock holdings. This increase has generated considerable interest among policymakers in identifying the magnitude of the stock market wealth effect on consumption. For example, in his recent testimony before the U.S. Congress, Chairman Alan Greenspan has stated that wealth-induced consumption growth has partly been responsible for generating aggregate demand in excess of potential. The Chairman says that the wealth effect may have added on average 1 1/2 to 2 percentage points to annual growth rate of real GDP over the last few years.3 The empirical work here addresses the wealth effect by considering aggregate consumption equations that relate consumption directly to equity wealth.

Poterba (2000) has suggested that the marginal propensity to consume out of wealth may have declined in the 1990s. According to Poterba, the main reason for this decline is the growing importance of equities in household wealth. Since the number of households holding equities is still lower than the number holding many other kinds of assets, and since a growing part of equity investments are held in tax-favored retirement accounts, the marginal propensity to consume out of equity wealth may be small. Furthermore, the marginal propensity to consume out of total wealth may appear to decline if the recent increase in household wealth reflects the growing importance of equities, as has been the case in the 1990s.4 In order to see whether the relationship between consumption and wealth has changed during the 1990s, I estimate consumption equations over the full sample period 1959Q1 to 2000Q2 as well as two other subperiods ending in the early 1990s, 1959Q1 to 1990Q2 and 1959Q1 to 1995Q2.

The empirical results that are presented here show that aggregate consumer spending is cointegrated with labor income and wealth over the sample period 1959Q1 to 2000Q2, indicating the existence of a long-term equilibrium relationship between consumption and its economic determinants, such as income and wealth. The coefficients that appear on income and wealth variables in the estimated cointegrating regression are statistically significant and measure the long-term responses of consumption to income and wealth. The results thus indicate wealth has a significant effect on consumption. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.